BMC Evolutionary Biology BioMed Central

Research article Open Access DNA evidence for global dispersal and probable endemicity of protozoa David Bass*1, Thomas A Richards1,2,3, Lena Matthai1, Victoria Marsh1 and Thomas Cavalier-Smith1

Address: 1Department of Zoology, The University of Oxford, South Parks Road, Oxford, OX1 3PS, UK, 2Department of Zoology, The Natural History Museum, Cromwell Road, London, SW7 5BD, UK and 3School of Biosciences, Geoffrey Pope Building, University of Exeter, Streatham Campus Exeter, EX4 4QD, UK Email: David Bass* - [email protected]; Thomas A Richards - [email protected]; Lena Matthai - [email protected]; Victoria Marsh - [email protected]; Thomas Cavalier-Smith - [email protected] * Corresponding author

Published: 13 September 2007 Received: 26 February 2007 Accepted: 13 September 2007 BMC Evolutionary Biology 2007, 7:162 doi:10.1186/1471-2148-7-162 This article is available from: http://www.biomedcentral.com/1471-2148/7/162 © 2007 Bass et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract Background: It is much debated whether microbes are easily dispersed globally or whether they, like many macro-organisms, have historical biogeographies. The ubiquitous dispersal hypothesis states that microbes are so numerous and so easily dispersed worldwide that all should be globally distributed and found wherever growing conditions suit them. This has been broadly upheld for protists (microbial eukaryotes) by most morphological and some molecular analyses. However, morphology and most previously used evolutionary markers evolve too slowly to test this important hypothesis adequately. Results: Here we use a fast-evolving marker (ITS1 rDNA) to map global diversity and distribution of three different clades of cercomonad Protozoa (Eocercomonas and Paracercomonas: phylum Cercozoa) by sequencing multiple environmental gene libraries constructed from 47–80 globally- dispersed samples per group. Even with this enhanced resolution, identical ITS sequences (ITS- types) were retrieved from widely separated sites and on all continents for several genotypes, implying relatively rapid global dispersal. Some identical ITS-types were even recovered from both marine and non-marine samples, habitats that generally harbour significantly different protist communities. Conversely, other ITS-types had either patchy or restricted distributions. Conclusion: Our results strongly suggest that geographic dispersal in macro-organisms and microbes is not fundamentally different: some taxa show restricted and/or patchy distributions while others are clearly cosmopolitan. These results are concordant with the 'moderate endemicity model' of microbial biogeography. Rare or continentally endemic microbes may be ecologically significant and potentially of conservational concern. We also demonstrate that strains with identical 18S but different ITS1 rDNA sequences can differ significantly in terms of morphological and important physiological characteristics, providing strong additional support for global protist biodiversity being significantly higher than previously thought.

Page 1 of 13 (page number not for citation purposes) BMC Evolutionary Biology 2007, 7:162 http://www.biomedcentral.com/1471-2148/7/162

Background what level of taxonomic distinction are protists in fact cos- In 1934, Baas-Becking stated with respect to microbes that mopolitan [22]. Foissner and colleagues promote a mod- 'everything is everywhere' with the important qualifica- erate endemicity model [23,24], which proposes that tion 'the environment selects' [1]. This ubiquitous disper- while some protist morphospecies have cosmopolitan sal hypothesis (UDH) of global dispersal with transient distributions, other, perhaps rarer protists have restricted local occurrence is the focus of increasingly active debate distributions. Evidence for this hypothesis is comprehen- [2-5]. It is supported by many surveys of protist morpho- sively reviewed in [25]. Our culture-independent, molec- species on a global scale, but exceptions have been pro- ular screening approach provides a robust framework for posed [6-8]. Microbial cosmopolitanism is thought to be testing the moderate endemicity model at high phyloge- driven primarily by the random dispersal pressure gener- netic resolution, as it does not distinguish between pro- ated by vast population sizes of cells below a body-size of tists which differ in ease of propagation in the laboratory, approximately 1 mm [9]. A dependent hypothesis states or ease of detection and study by microscopy. that global protist species richness is consequently rela- tively low, as opportunities for allopatric divergence are Cercomonads are benthic heterotrophic flagellates (5–50 limited or absent, and that the local to global species ratio µm long) comprising the genera Cercomonas, Eocer- is high, theoretically approaching 1 [5]. In other words, comonas, and Paracercomonas [14]. They are abundant in the diversity of microbial species found in one locality soils [26-28] and aquatic environments, being among the generally represents global levels of diversity within that most prominent bacterivorous protozoa. As global disper- habitat. However, recent molecular studies have contrib- sal should theoretically be easier for marine protists, mak- uted several new perspectives [7,10-16], mainly 1) that ing endemicity less likely, we preferentially sampled soil diversity of 18S rDNA genotypes (18S-types) within mor- and freshwater environments to test the UDH more rigor- phospecies can be very high; 2) that many 18S-types have ously. As some cercomonad 18S-types are globally distrib- cosmopolitan distributions, and 3) some 18S-types uted [10,14], we used ITS1 rDNA sequences, which evolve within morphospecies have ecologically and/or geograph- significantly faster than 18S rDNA sequence (Fig. 1), to ically restricted distributions. These and other investiga- determine whether geographically restricted distributions tions indicate that protist morphospecies usually offer too could be detected within cercomonad 18S-types at this crude an evolutionary resolution to be ecologically higher level of resolution. As there are hundreds of differ- informative [17-21], reinvigorating the debate about at ent cercomonad 18S-types [14], we used PCR primers tar-

NZ1.5C New Zealand XT160 UK XT179 New Zealand clade α - EC South Africa Paracercomonas 5A New Zealand P111 Panama Panama 61 Panama 107 β Panama 101 clade - Paracercomonas Panama 89 Panama 86 C73 Spitzbergen 11.7E Spain clade γ - C76 Antarctica Eocercomonas XT196 UK SSU rDNA ITS rDNA } 0.1 subs/site

ComparisonsFigure 1 of evolutionary changes in the 18S and the ITS1 sequence Comparisons of evolutionary changes in the 18S and the ITS1 sequence. Left tree (black): Maximum likelihood (ML) phylogeny of cercomonad 18S rDNA sequences calculated from an alignment of 15 sequences and 213 nucleotide positions from the hypervariable V4 region. Right tree (grey): ML phylogeny of corresponding ITS1 rDNA (213 positions). Branch lengths and scale bars standardised to illustrate the much greater evolutionary rate of ITS1 than even the most hypervariable 18S rDNA region. Equivalent parsimony analyses using gaps as a 5th character (not shown) showed a parsimony score of 367 (minimum changes) for ITS1 and 120 for the V4 18S rDNA region. Cultures used for ecological and phenotypic character experiments (Table 2) marked by circles.

Page 2 of 13 (page number not for citation purposes) BMC Evolutionary Biology 2007, 7:162 http://www.biomedcentral.com/1471-2148/7/162

geted to three very narrow taxonomic groups comprising lades of α (Fig. 3B) were designed in order to probe inten- 1 to 3 18S-types each (Figs 2, 3). This approach enabled us sively previously untested eDNAs to determine whether to maximize the chances of detecting the same genotypes ITS types retrieved from only one or two locations in the in different samples even when rare. To be representative, original analysis were in fact more widely distributed. we targeted three different regions of the cercomonad tree PCR screening and sequencing showed that clade a (Fig. incorporating at least one genus from each of the two 3B), previously detected only from Peru was also present main cercomonad clades (Fig. 2) [14]. Our intensive sam- in New Zealand, although clade f, originally from Aus- pling of the rapidly-evolving ITS1 rDNA marker reveals tralia/New Zealand could still only be found in Australa- both ubiquitous and restricted distributions and thus sian samples. These additional experiments show that demonstrates that biogeography in microbes and macro- there was at least some undersampling in our initial data organisms differs only in degree. We also show that cer- set. Thus some ITS-types are probably actually more comonad strains within a single 18S-type can differ in widely distributed than Figs 3B–D suggests, especially for terms of phenotype and ecological preferences. groups β and γ where this more intensive additional sam- pling was not done. However, this extra sampling of Results and discussion group α increased the geographic distribution for only Fig. 2 is a general tree of cercomonad diversity, showing two clades. the positions of the three very narrow lineages targeted in this study. In order to achieve a very intense sampling of Furthermore, of 21 ITS-types detected from only one bio- these lineages and avoid amplifying a vast number of geographical region (grey arrowheads, Fig. 3B–D), five intermediate lineages (much greater than shown on this were retrieved from more than one library, all within sample tree) irrelevant to our present purpose we used group α. Given the high level of our sampling, this sug- PCR primer pairs including one primer highly specific for gests that these ITS-types are biogeographically restricted each narrow lineage (see Methods). Fig. 3 shows that (our additional sampling and lineage-specific primers despite the extreme narrowness of our primer specificity found none of these ITS-types elsewhere). It is particularly we recovered up to 18 distinct ITS-types within a single striking that sequence DB-0505-31 on Fig. 3B was 18S-type, and that two groups (β,γ – Fig. 3C,D) have some detected in three separate freshwater Australasian libraries globally distributed ITS-types (six found in both hemi- and in no others. One should therefore not extrapolate spheres and three or more biogeographical regions, treat- from our finding four clear examples of global ubiquity to ing the Palaearctic as a single region). The other group (α the conclusion that this is true of all cercomonad ITS- – Fig. 3B) was sampled from only three biogeographical types. It cannot reasonably be argued that all genotypes regions (Table 1), and revealed many ITS-types from two are both globally distributed and equally abundant on all regions but only one from all three. Most ITS-types (21 of continents. If they were, we should easily have found DB- 36) were detected only in one region. The provenances of 0505-31 in libraries constructed from other continents. It all 18S- and ITS-types are shown in Table 1. As more seems probable that it is either really restricted to Austral- libraries were screened within each clade, individual ITS- asia or else dramatically more abundant there at the time types were increasingly recovered from more than one of sampling. Even the latter is contrary to the core assump- continent. This suggested that the true diversity present in tions of the UDH, unless it could be demonstrated that the environment remained undersampled. To explore this none of the samples on any other continents were from further, we did two things. ecologically similar sites to those sampled from Australa- sia. It is possible that such ITS-types have a patchier distri- Firstly, for group α, which contains the largest number of bution than those that are easily found at many sites putatively geographically restricted ITS-types, more clones globally; but although patchiness combined with rarity were sequenced from the libraries constructed from the and undersampling might account for the restricted distri- largest number of successfully amplified eDNAs (environ- bution of some of the sequences we encountered in single mental DNA samples; see Methods) – those from Europe libraries, it is unlikely to explain all examples of apparent marine and Panama marine – and those producing the biogeographic restriction. Three of the five ITS-types highest number of putatively 'endemic' (and incidentally detected in more than one library but from only one bio- the most divergent) ITS-types – New Zealand and Aus- geographical region were from Australasia. These are tralia; see Table 1. All 44 new sequences from these four shown in Figs 3B and 3C as clusters X (including DB- libraries matched ITS-types previously recovered; suggest- 0505-31 mentioned above) and Z. Both putative clades ing that sampling of these libraries was at or near satura- are supported by multiple unique and identical sequence tion (16 ITS-types being recovered from 110 sequence motifs in conserved alignment regions of ITS1, as shown reads). One of these ITS-types (h, Fig. 3B), previously only in Fig. 4. X and Z are illustrative of the dominance of Aus- recovered from Panama, was thereby shown to be present tralasian-derived sequences (12 ITS-types) in the set of 21 in European samples. Secondly, primers specific to subc- ITS-types recovered from only one biogeographical region

Page 3 of 13 (page number not for citation purposes) BMC Evolutionary Biology 2007, 7:162 http://www.biomedcentral.com/1471-2148/7/162

Cercomonas longicauda CCAP1910/2 DQ442884

18.6E (Oxford) AY884328 100 Cercomonas sp. ’large’ AF411266 75 NZ1.7E (New Zealand) AY884322

C-43 (Mylnikov) AY884334 Cercomonas 100 C-85 (Mylnikov) AY884325

NY-1 (Mylnikov) AY884320

Environmental 4_2.2 (UK moss) AY884313

Environmental 7_3.6 (UK marine) AY884317 100 11.7E (Spain) 100 C76 (Antarctica) γ

C73 (Spitzbergen) Eocercomonas Eocercomonas ramosa AY884327 99 Eocercomonas sp. E AF411269

Environmental 13_3.5 (Canada soil) AY884316

NZ1.5C (New Zealand) α 100

100 Paracercomonas marina ATCC 50344 AF411270

100 19.5C (Malaysia) AY884341 100 RS/18A ATCC 50316 U42448

C-71 (Mylnikov) AY884340 82 Panama 89 Paracercomonas β 100 Panama 101

100 Panama 61 & 107

Cs-9 (Mylnikov) AY884337 ’Tempisque’ 97 AF411271 cercomonad anaerobic sp. Clade 2B ATCC 50367 AF411272 N-Por filose amoeba AF289081

82 Massisteria marina AF174372 Proteomyxidea 100 Dimorpha-like

Gymnophrys cometa AJ514866 0.01 subs/site

PhylogenyFigure 2 of cercomonads Phylogeny of cercomonads. BioNJ bootstrap tree (1000 replicates; corrected for Γ & I) of a subset of known cercomonad lineages, representing all known main cercomonad clades and genera (Cercomonas, Eocercomonas, and Paracercomonas) as defined by Karpov et al. [14]. The composition and phylogenetic position of the groups α, β, and γ forming the basis of this study are shown. A group-specific primer set was designed for each of these groups (see Methods).

Page 4 of 13 (page number not for citation purposes) BMC Evolutionary Biology 2007, 7:162 http://www.biomedcentral.com/1471-2148/7/162

GlobalFigure provenance, 3 biogeography and phylogeny of cercomonads Global provenance, biogeography and phylogeny of cercomonads. (A) World map showing major animal biogeo- graphical regions. (B-D) 18S-ITS1 rDNA phylogenies of groups α (278 nucleotide positions), β (414 positions), and γ (335 posi- tions). See Methods for details of each sequence dataset. Coloured discs correspond to hoops on map. Grey arrowheads indicate sequences found in only one biogeographical region. Square black brackets indicate 18S-genotype boundaries. X and Z (bracketed) indicates a possible endemic genotypic radiation. Branches marked a, f, and h in (B) are discussed in relation to additional sampling and ITS-type-specific primer probing. Branches without coloured discs originate from cultures. Sampling range differs among the three groups (Table 1). Key to bootstraps (ML/parsimony) given to right of scale. Bootstraps shown only when equal to or exceeding the value stated, although precise values are shown for clades marked X and Z, as these are discussed in the text as potentially geographically-restricted clades. Branch labels underlined and in bold refer to culture- derived sequences, including those used in the culture experiments. NB: in some cases a single branch corresponds to >1 cul- ture or both environmental and culture-derived sequences. The long branch for 11.7E is not shown in (D) for the sake of clar- ity. It was not detected in any environmental library.

Page 5 of 13 (page number not for citation purposes) BMC Evolutionary Biology 2007, 7:162 http://www.biomedcentral.com/1471-2148/7/162 Table 1:Sampling andproven tipa ot fiaSi 12(+2 11 Soil Africa South Ethiopian ertcB ol615006&09.02 0 12.04 05.02 5 0 1 1 2 5 1 1 6 5 11 Soil Freshwater sediment BC Coastal sediment Japan Nearctic UK E. Palaearctic W. Palaearctic etoia euSi 3 07.05 07.03 2 2 02.03 5 4 6 1 1 4 7 (3) 7 (4) 1 8(4) Soil Soil Oxfordshire W. Palaearctic screened #eDNAs Peru Soil&f/w sediment Neotropical NewZealand Australasian GROUP Habitat Provenance Biogeog. region ertcB osa eiet511006&09.02 02.03 02.05 12.04 0 & '04 '03 3 03.03 0 0 02.03 1 1 5 (1 0 1 1 1 2 4 1 1 2 1 4 1 1 5 6 8 (1) (eDNA =environmental DNAsample –see Methods for moredetails.) 3 4 (3) 6 (5) Soil&f/w sediment Panama 6 7 Coastal sediment Soil&f/w sediment Neotropical NewZealand Freshwater sediment Australasian BC 8 GROUP Japan Nearctic Coastalsediment Freshwater sediment Soil &Europe UK E. Palaearctic Palaearctic Calcutta W. Panama Neotropical Soil&f/w sediment Oriental NewZealand Australasian GROUP These additional genotypeswere alsodetected, andwere α γ β aaaCatlsdmn 03.03 2 1 7 Coastal sediment Panama xodhr ol41507.03 5 1 4 Soil Oxfordshire KCatlsdmn 7 05.02 12.04 12.04 03.03 03.03 4 4 4 2 1 4 1 1 1 1 8(7) 4 (1) 4(1) 5 (3) 4 (4) Coastal sediment Coastal sediment UK Freshwater sediment Panama Soil Panama Soil Tasmania Australia utai ol41412.04 07.05 & '04 '03 03.03 12.04 4 1 1 1 1 1 4 1 1 1 1 12 7 Soil 4 (1) 8 Australia Soil Soil &Europe UK Coastal sediment Peru Panama Soil Tasmania ance ofgenotypessampled osa eiet51202.03 02.03 2 1 1 5 1 Coastal sediment 13 Coastal sediment (# successful amplifications, where known) where amplifications, (# successful notfrom any recovered other site.(Not shown on Fig. 3.) 1Stps#T-ye #ITS-types found #ITS-types #18S-types 1 )(1 1 )4 (+2 1 )(1 1 )2 (+2 only in this biogeog. region biogeog. this in only 1 )03.03 1 )12.03 Date of collection (MM.YY)

Page 6 of 13 (page number not for citation purposes) BMC Evolutionary Biology 2007, 7:162 http://www.biomedcentral.com/1471-2148/7/162

A. 117 142 197 220 DB-1002-6 CCGCGTAGTTGTGTTGGCGCAAAG GGCTATCATGGCTGTGTTTTTCC DB-0505-30 CCGCGTAGTTGTGATGTCGTAAAG GGACCTCACGGCTGTGTTTTTCC DB-0505-41 CCGCGTAGTTGTGTTGTCGCAAAG GGACGTCATGGCTGTGTTTTTCC DB-1602-7 / Panama 111 CCGCGTAGTTGTGTTGTCGCAAAG GGACATCACGGCTGTGTTTTTCC DB-0802-4 CCGCGTAGTTGTGTTGTCGCAAAG GGACACTACGGCTGTGTTTTTCC DB-1602-28 CCGCGTAGTTGTAGCGCCGCAAAG GGACGCTATGGCTGTGTTTTTCC DB-0505-31 CCGCGTAGTTGTAGCGCCGCAAAG GGACGCTATGGCTGTGTTTTTCC X DB-0802-29 CCGCATAA-TGTGGCGTCGCAAAG GGATGCCATCATTGTGTTTTTCC DB-0802-7 CCGCAT-GATGCAGTATCGCAAAG GGATGCTGTCATTGTGTTTTTCC DB-1602-6 CCGCATA-TTGCAAAATCGCAAAG GGATTCTGTAG-TGTGTTTTTCC XT160/XT179/NZ1.5C CCGCGTAATTGTCACGTCGCAAAG GGACGCAACAATTGTGTTTTTCC DB-1602-2 CCGCATGGTTGTAGTATCGCAAAA GGATGCTGTGACGATGTTTTTCC EC South Africa /5A New Zealand CCGCATGGTTGTAGTATCGCAAAG GGATGCTGTGACGATGTTTTTCC DB-2102-7 CCGCATGGTTGTAGTATCGCAAAG GGATGCTGTGACAATGTTTTTCC DB-0802-15 CCGCATGGTTGTAGTATCGTAAAG GGATGCTGTGACAATGTTTTTCC DB-0505-3 CCGCATGGTTGTAGCGTCGCAAAG GGATGCTGTGACAATGTTTTTCC DB-0505-42 CCGCATGGTTGTAGCGTCGCAAAG GGATGCTGTGACAATGTTTTTCC DB-1602-30 CCGCATTATTGTAGTATCGCAAAG GGATGCTATGGCAATGTTTTTCC 0.05 subs/site

B. 210 224 325 345 363 372 4-28.3/Panama 101 GCAAGGACCGTCCT GATCATTAAACAAATCCACG TTATTT---TGG 14-11.4 GCAAGGACCGTCCT GATCATTAAACAAATCCACG TTATTT---TGG 23-11.4-E GCAAGGACCGTCTT GATCATTAAACAAATCCACG TTATTT---TGG Panama 89 & 86 GCAAGGACCGTCCT GATCATTAAACAAATCCACG TTATTT---TGG 15-11.4 ACAAGGACCGTCCT GATCATTAAACAAATCCACG TTATTT---TGG ACAAGGACTGTCCT GATCATTAAACATATCCATG TTAAGGACGAGG 16-11.4-NZ Z 9-NZ TCAAGGACTGTCCT AATCATTAAACATATCCAAG TTTTTGA-GTGG Panama 107 & 61 ACAAGGACCGCCCT GATCATTAAACAAATCCACG TTCGTTAATTGG 0.01 subs/site

FigureShared alignment4 characters supporting putative geographically restricted clades X and Z (boxed in red) Shared alignment characters supporting putative geographically restricted clades X and Z (boxed in red). (A) Paracercomonas group α tree; (B) Paracercomonas group β tree (taken from Fig. 3). Characters at specific alignment positions are coloured black when 0.8 of positions agree; otherwise they are unshaded. Unique clade-specific alignment characters are illustrated with triangles across the bottom of the alignments. Shared sequence motifs in a hyper-variable section are illustrated with an oval across the bottom. Sequence position numbers across the top refer to DB-1002-6 and P101 PANAMA.

(= 57%). As is well known, Australasia is the most geo- marine transition is unusually easy and rapid for these cer- graphically isolated continent and shows exceptionally comonads. Protozoa in marine and non-marine habitats high endemicity for higher animals, e.g. monotremes and are largely distinct [11,12]. Many major taxa are restricted marsupials, and plants [29,30]. Perhaps it also has an to one or the other environment: Radiozoa, Phaeodarea, unusually high level of endemic cercomonad genotypes. Xenophyophorea, Pelagophyceae, and most foraminif- eran subgroups are exclusively marine, whereas Myceto- It is theoretically possible that seasonal differences in cer- zoa (for example) are never marine. Even in groups where comonad strain occurrence and/or relative abundance some marine/freshwater shifts have occurred, there are could account for some of the distribution patterns many substantial subclades that are apparently exclusively described here. However, unpublished data from compre- one or the other, for example in bodonids and Goni- hensive sampling of cercomonads at several sites in the omonas [11,12,15]. Some lineages have clearly undergone UK suggest that if such shifts in abundance occur they such shifts more frequently than others and are appar- would certainly not preclude detection by the very specific ently more adaptable in this respect. Yet even in such primers used in this study (Howe and Bass, unpublished). groups not all strains can be encouraged to adapt physio- Many, if not all, genotypes detected at these sites were logically from 'marine' to 'non-marine' (or vice versa) con- present in both winter and summer. The times of sample ditions in laboratory cultures, and even where this can be collection for the present study are given in Table 1. achieved there is no guarantee that it can and does happen in nature. On present evidence the fraction of protist line- Nine ITS-types were recovered from both marine and non- ages able to live in both marine and non-marine habitats marine environments (Fig. 3B–D). Though only a small is very small. proportion of the total, this suggests that the marine/non-

Page 7 of 13 (page number not for citation purposes) BMC Evolutionary Biology 2007, 7:162 http://www.biomedcentral.com/1471-2148/7/162

The unusual capacity of some cercomonads to make rela- – Table 2) they are 'formed readily' [31], whereas this is tively rapid switches between marine and non-marine not true of all other ITS-types within that 18S-type (Table environments might greatly facilitate their global disper- 2). Our culture experiments also showed that cultures of sal. To see if an ability to grow in marine habitats is a gen- identical ITS-types were more similar to each other than eral property of these cercomonads, we tested strains from those with different ITS1 sequences in respect of a suite of clades α and γ isolated from soil or freshwater in marine traits, which are likely to be ecologically significant (Table media: strains XT179, NZ1.5C, XT160, EC South Africa, 2). 5A New Zealand, and Panama 111 (all clade α), and C73, XT196, 11.7E, and C76 (clade γ) were inoculated into These observations suggest that differences in ITS media at different concentrations, and observed for five sequence may be useful as markers for identifying clearly weeks. The results are summarized in Table 2. None of the distinct biological/taxonomic units. It has been shown for strains could grow in full strength seawater medium strains from a broad range of taxa (ciliates, green algae, (SWM), showing that not all strains with the same 18S plants, sea urchins, and abalone) that boundaries of sex- rDNA sequence can grow in both marine and non-marine ual compatibility appear to coincide with phylogenetic media, even though some ITS genotypes can be found in groups defined by individual compensatory base changes both habitats (Fig. 3). In fact, only strain 11.7E grew at all in ITS2 rDNA [32,33]. More recently, Amato et al. [34] well in 50% SWM, although the cells were unusually inac- have shown that cryptic sister species are profuse in the tive. All other strains tested grew very poorly or not at all diatom genus Pseudo-nitzschia, where biological species in 50% SWM, and grew with highly varying degrees of vig- correspond to strains with identical helix regions of ITS2. our in 25% SWM. Our observations, based on replicated, In Pseudo-nitzschia the degree of differentiation between standardized cultures, show that cercomonad strains with strains was similar for ITS2 and ITS1, although in at least identical 18S rDNA but different ITS1 sequences can differ one case divergence in ITS1 sequences was more marked in several respects: salinity tolerance, propensity to form than that in ITS2 [34]. With this one exception, ITS1 cysts, vigour in standard non-saline media (growth den- sequences within a biological species were identical. If cer- sity and persistence over time), size and basic cell mor- comonads are similar to diatoms, ciliates, and green algae phology (Table 2). In our experiments the greatest in this respect, then the majority of 'species' would have differences in salinity tolerance were between strains with no detectable variation in ITS1, and the majority of sepa- slightly different 18S sequences, even when the degree of rate ITS1 types would be separate species. In this scenario genetic difference was at or below our threshold for defin- it would be unsurprising that distinct ITS1-types respond ing unique 18S-types. This reinforces the growing aware- differently to culture variations. However, it is not known ness that taxonomic units based on 18S rDNA identity are whether cercomonads are sexual, and there are insuffi- at least sometimes insufficiently resolving to identify dis- cient morphological characters (without using electron tinct ecological entities [18-21]. Given the very close microscopy) that could be robustly related to ITS diver- rDNA similarity between the strains studied, it appears gence. Thus biological species cannot currently be defined that the evolution of many physiologically and evolution- for cercomonads. arily significant characters may be relatively rapid in cer- comonads. Conclusion Our study is so far unique in screening such large numbers The strains used for the culture experiments were all iso- of globally distributed eDNAs with PCR primers so nar- lated from soil and freshwater and we had no prior evi- rowly targeted as likely to include relatively few 'sibling dence suggesting that any could grow in marine media; species'. Our data demonstrate on the one hand that glo- none of these cultured ITS-types correspond to those bal dispersal of at least some cercomonad ITS-types occurs retrieved from the libraries (also indicative of non-exhaus- faster than divergences among these relatively fast-evolv- tive sampling of ITS-types). The culture experiments sug- ing markers. On the other hand, many ITS-types appear to gest that some ITS-types are tolerant to a wider range of be more restricted in distribution; several are candidates salinity levels than others, and that there are ITS-types for being genuinely endemic to one biogeographic region. which are genuinely restricted to non-marine environ- Still more intensive sampling would be necessary to ments. Of the 19 ITS-types detected in more than one decide whether the majority of strains found only once in library, none was exclusively marine, whereas nine were this study fall in the cosmopolitan or endemic category. found only in non-marine libraries. This suggests a fresh- However, our results are more decisive than earlier indica- water ancestry for these cercomonad groups and that only tions in the ciliate Paramecium that one sibling species of a minority of ITS-types tolerate high salinity. Many pro- the P. aurelia complex may be endemic to Eurasia [35]), tists persist in unfavourable environments by encysting. whereas many are cosmopolitan. Some assertions of Different ITS-types may also differ in propensity to form endemicity in Paramecium [35] were attributable to under- cysts; for example in Paracercomonas ekelundi (SCCAP_C1 sampling [36]. Several species are almost certainly genu-

Page 8 of 13 (page number not for citation purposes) BMC Evolutionary Biology 2007, 7:162 http://www.biomedcentral.com/1471-2148/7/162

Table 2: Results of culture experiments: ability of clonal cercomonad strains to grow in various marine media and other strain characteristics

Cladea Strain 18S identityb ITS1 identityb Morph. identityc Salinity Size (µm)e Cystsf Vigourg toleranced

α XT179 1 1 1 5–7 99()9 (motile cells) 9 9 9 9 9 α NZ1.5C 1 1 1 5–6 9 9 9 9 α XT160 1 1 1 5–7 9 9 9 9 9 9 9 α EC 1 * 2 * 1 5–7 X? 99 (immobile 9 cells) 9 9 α 5A 1 2 * 2 N/A 6–8 X? 9 9 α P111 1 3 N/A N/A 5–7 X 9 α SCCAP C1h 1 N/A N/A N/A 4–10h N/A 99()9 h

γ C73 1 1 N/A 4–6 X 9 9 9 γ XT196 1 2 N/A N/A 5–10 N/A 9 γ C76 2 3 N/A X 5 X 9 γ 11.7E 3 4 N/A 5–7 9 9 9 9 9 9 9

N/A = no data * Although the ITS1 sequences for EC and 5A are identical, and 18S rDNA is identical in the V4 region (on which Fig. 1 is based), EC has an insert near the 3' end of 18S and is therefore clearly a different strain. This is the only strain in this study known to have such an insert. a see Figs 1, 2, and 3. b Identical numbers indicate strains with identical sequences. c Identical numbers indicate strains with strain traits other than size (flagella length:body length, form and frequency of plasmodium formation, propensity to form cell clusters, etc.) which were indistinguishable from others in the same medium under light microscopy (200× phase contrast). d Overall performance of strains under a range of saline medium regimens, as described in Methods. X means that no living cells were seen. e Size range determined from 3+ independent cultures observed over a period of 18 months. f Indicates the propensity of the strain to form cysts in Volvic & grain cultures, summarized from observations of 3+ independent cultures observed over a period of 18 months. X = no cysts seen, 9 = cysts infrequent and sparse, 99 = cysts regular at low to medium densities, 999 = cysts readily formed and seen at high densities. g A measure of the ability of the strain to form dense cultures in Volvic & grain medium, which persist well over time (2+ months in unmanipulated cultures), summarized from observations of 3+ independent cultures observed over a period of 18 months. 9 = rarely or never reach high density, 99 = may reach medium to high density but do not persist, 999 = usually reach high densities and/or persist for long periods without encysting or disappearing from the culture. h See ref. 31. Not included on Figs 1 and 3 as there is no ITS1 sequence available. Not included in culture experiments.

Page 9 of 13 (page number not for citation purposes) BMC Evolutionary Biology 2007, 7:162 http://www.biomedcentral.com/1471-2148/7/162

inely geographically restricted (i.e. not cosmopolitan; ing material associated with mosses, freshwater sediments [35,37]), though in some cases apparently because of lat- and scrapings from submerged rocks, plants, and detritus, itudinal/climatic factors [35] rather than limited dispersi- and coastal sediments and rock scrapings. bility, and thus not counter-examples to the ubiquitous dispersal theory. Targeted species-specific environmental Molecular Screening DNA sampling is needed in these ciliates along the lines Cercomonad clade-specific primers were designed from of our present study to distinguish between undersam- culture-derived 18S rDNA alignments: forward primers pling, latitudinal/climatic geographic restriction, and gen- VM1B (group α) 5'-TAGTACAACGTAACCCTTGGT, uine historically based endemicity. Thus, just as for VM4B (group β) 5'-AATAGTGGGGCTTAGGCTTGCC and animals and plants, some protist species or reasonable VM5 (group γ) 5'-CTTAGTGAGCTTCAGAGATTGATGTA- surrogate markers for species such as ITS genotypes are CATA. Reverse primers, situated in 5.8S rDNA, were the genuinely cosmopolitan, whereas others show continen- eukaryote-universal 369R 5'-TCGCATTACGTATCGCATT- tal endemism. In this respect our results support the mod- TCGCTG (first round PCR) and Bii 5'-TGCGTTCT- erate endemicity model described above. TCATCGWTACGAG (nested PCR). The primers were designed from a total cercomonad alignment of approxi- To measure true levels of endemism in microbial eukary- mately 150 sequences to be specific to one or more cer- otes would require extensive testing of a large number of comonad 18S-types. The phylogenetic positions of the protist groups from across the eukaryote tree with at least three groups is shown in Fig. 2 in the context of a subset as much intensity as we demonstrate in this paper. What of all cercomonad lineages. The exact location of the is now clear is that some of the perception of greater cos- primer site in the SSU was determined by the existence of mopolitanism in microbes [5] is attributable to the taxo- regions that provided the degree of specificity required. nomic artefact of lumping huge numbers of genetically This degree of specificity required between groups: α com- very different organisms into a single crude 'morphospe- prises a single 18S-type because it was relatively frequently cies'. Our study confirms that the 18S-type, currently the represented in our database, and therefore we judged that most frequently used marker for distinguishing protist it would be relatively easy to find in eDNAs. β comprises strains, is too coarse for distinguishing between ecologi- three, quite divergent, 18S-types – the interest here was cally and biogeographically significant units in cercomon- that we had only previously retrieved members of this ads. The next key question is whether cercomonad ITS- clade as cultures from Panama. γ comprises three very types correspond to biological species. If they do, cer- closely related 18S-types, individually for which it would comonad species numbers are likely to be at least 100 have been effectively impossible to design reliably robust times greater than yet described. Thus currently the primers. All primer sets were tested with known positives number of protozoan species is probably hugely underes- and negatives, and a PCR protocol chosen that amplified timated. only sequences within those clades.

Methods Nesting was necessary as the initial amplifications fre- Environmental Sampling quently produced faint/invisible bands after gel electro- Globally dispersed environmental samples were collected phoresis. PCR cycling conditions were 5 min. initial from soil, freshwater, and marine sediments, by different denaturation (95°C), followed by 35 cycles of 32s at individuals using separate collecting apparatus, and DNA 95°C, 36s at 68°C (α), 65°C (β and γ), 1.5 min. at 72°C, extracted under sterile conditions as described previously then 5 min. at 72°C. Multiple PCR products for each envi- [10]. Provenances of all environmental DNA extractions ronment were pooled prior to separation on a 1% agarose (eDNAs) used in this study are shown in Table 1. In each gel from which they were excised and cleaned using the sampling region a range of habitats was sampled over as GFX Gel Purification Kit (GE Healthcare). The resulting broad a geographical area as possible. A broad overview of fragments were cloned using the TOPO TA Cloning Kit sites is as follows. Panama: Pacific and Caribbean slopes (Invitrogen). At least 16 colonies were selected from each (coast and interior); New Zealand: North and South library and sequenced as previously reported [10]. Island (coast and interior); Peru: soils from both sides of the Andes – predominantly lowland wet tropical forest; Sequences of each group were aligned manually using SE- Tasmania: soil samples from around the perimeter of the AL [38], and preliminary BioNJ trees constructed [39]. island; British Columbia: a transect from Vancouver to the These were used to refine the environmental sequence Okanagan valley, plus the West coast of Vancouver Island; alignment and reduce the environmental sequence sam- Europe: primarily UK but also France, Germany, and pling (113 down to 18 group α sequences; 110 to 9 group Greece. Less widespread samples were taken in India, β; 109 to 16 group γ) and determine genotype boundaries Japan, and Australia. The habitats sampled included wet as described below. All effectively identical sequence cop- and dry soils, scrapings from rocks, sediments and decay- ies were then removed from the alignment, leaving only

Page 10 of 13 (page number not for citation purposes) BMC Evolutionary Biology 2007, 7:162 http://www.biomedcentral.com/1471-2148/7/162

those sequences deemed to be clearly different from each amplification. The corresponding primer sequences are other. The length of the amplicon for each group was a1: 5'-ATTATCAGTTAAAGGCTTTCGCCTAG; a2: 5'-ACA- dependent on the position of the group-specific (forward) GAATCCCTCATCGCTAGGATG (clade a, originally primers in the 18S rDNA, at the 5' end. The full amplicon detected only from Peru); f (clade f, originally detected length was retained in the alignments for Figs 3B–D in only from New Zealand and Australia): 5'-AGAGA- order to maximize tree quality. The alignment lengths GAAGCGGGCCATAACAC. Following PCR optimization analyzed were as follows: group α = 278 positions; group of each primer set using known positives and negatives, β = 414 positions; group γ = 335 positions. Final trees primer sets a and f were each used to probe eight eDNAs were calculated using maximum likelihood (ML) with an from four locations: Australia, New Zealand, Tasmania, eight category Γ + I site rate variation model. All model and Panama, including the eDNA from which bands were ML parameters were previously estimated using likeli- obtained using the VM1B-Bii set. The PCR products from hood treescore from a preliminary BIONJ tree [39,40]. We each location were pooled separately and run on a gel. used a general time-reversible rate matrix (GTR) with six Primer f produced a band only for Australasian region substitution categories estimated from the alignment samples, whereas primer a2 (following nesting) produced using the BioNJ tree. Additional analyses was performed bands for both Australia and Panama (no bands for either by maximum parsimony with gaps coded as a 5th charac- were seen after the first nested PCR using a1). The two vis- ter state [39]. In both approaches 10 random replicate ible bands were cloned and sequenced, yielding only the heuristic searches with stepwise addition and TBR were identical ITS-type to which they were designed. This sug- used for the topology searches (ML trees shown on Fig. 3). gests that a double nesting PCR approach (i.e. three serial 100/1000 bootstrap replicates (for ML/parsimony respec- amplifications in all) can be necessary when probing with tively) were conducted using the same settings. such specificity, probably because of the low concentra- tions in environmental samples of some individual ITS Evolutionary rate comparison (Fig. 1) types. To confirm that the cercomomad ITS sequences evolve significantly faster than any region of 18S rDNA sequence Estimation of clonal variation in rDNA cistrons and PCR/ we constructed complementary alignments for both data- sequencing error sets in SE-AL using all available sequences derived from As a control for properly interpreting whether observed cultured representatives of the three cercomonad groups differences in ITS could be caused by polymorphic differ- studied. We calculated the phylogenies using parsimony ences among the many copies within a single genome and ML as described above [39,40]. Scales of the ML and/or by PCR/sequencing errors, we made rDNA librar- results were standardised and differences in parsimony ies from clonal cultures and measured intra-clonal differ- scored (minimum changes) for each alignment were ences in the resulting alignment for two phylogenetically recorded. The length of both 18S and ITS1 alignments was distant cercomonad strains (Cercomonas sp. 'AZ6' ATCC 213 positions. The beginning of ITS1 was determined to 50418 &Cercomonas C59 [14]). For each strain 24 clones follow the end of the 18S sequence as given in [41], which were sequenced on both strains along the full amplicon point coincided with the junction between conserved length. The sequence difference percentage was calculated (18S) and highly variable (ITS1) regions in our extensive by dividing the total number of mismatches by (24 × cercomonad rDNA alignment. The V4 region of 18S rDNA number of alignment positions) then multiplying by 100. was selected as the most variable region of that gene, to The distribution of sequence variation in the alignment is provide the most appropriate comparison between the shown in Table 3, which shows that sequence variation genes and also because it is the region we use to define caused by combined PCR nucleotide incorporation error unique 18S-types, in a similar way to that described for and intra-clonal rDNA polymorphism is low (mean = ITS1, below. 0.1%) and that the more rapidly evolving regions of the gene are not more variable in this respect than relatively Group α subclade-specific primers conserved regions. It is not possible to distinguish An initial amplification was carried out with VM1B and between actual intra-nuclear differences between gene 369R. The products of each reaction were diluted 1:10 and copies and nucleotide misincorporation errors during the used in a nested PCR with VM1B and highly specific sub- initial PCR. However, this approach estimates or overesti- clade primers for the two subclades a and f on Fig. 3B. mates the upper limit of intra-nuclear polymorphism lev- These primers were situated in ITS1 rDNA to provide the els and provides a measure of the total intrinsic level of necessary high degree of specificity. The PCR protocols errors generated by our methodology. The alignments were optimimzed in the same way as described for α-, β-, were created from an initial PCR amplification of DNA γ-specific primers. For clade a, two specific primers were extracted from each culture using the cercozoan-specific used, a1 and a2, the latter being used in a second nested primer 1259F (5'-GGTCCRGACAYAGTRAGGATTGACA- PCR step using the diluted products of the VM1B-a1 GATTGAAG) and the eukaryote-universal primer 28Sr1

Page 11 of 13 (page number not for citation purposes) BMC Evolutionary Biology 2007, 7:162 http://www.biomedcentral.com/1471-2148/7/162

Table 3: Sequence variation within libraries constructed from two clonal cercomonad cultures

Strain Position on rRNA Number of Number of Sequence Variability of cistron positions examined mismatches difference (%) region

AZ61 18S 545 15 0.11 Relatively conserved C-591 535 11 0.09 AZ6 ITS1-5.8S-ITS2-28S 603 11 0.08 Relatively variable (D1/D2 region) C-59 609 20 0.14 AZ6 whole fragment 1148 26 0.09 (average) C-59 1144 31 0.11

1 see ref. 14

(5'-CGGTACTTGTTCGCTATCGG). The primers used for pendent libraries on multiple occasions is extremely sequencing were 1259F, 369R, pre-Bf (5'-GTAGGTGAAC- remote. CTGCAGAAGGATC) and 1733R (5'-TGATCAAGTTTGAT- TCAGTTCTCGGAT). Culturing To test whether monoclonal cercomonad strains isolated Definition of unique genotypes from non-marine environments can grow in marine The threshold of divergence used for defining unique gen- media, we inoculated strains from clades α and γ into the otypes was much higher than divergence due to combined following culture media: Volvic mineral water (Danone) errors, which were measured as described in the preceding enriched with a sterile (boiled) barley grain; 100% Artifi- section. The standard requirement for two genotypes to be cial Sea Water (ASW; CCAP recipe); 50% ASW; and 25% considered distinct from one another was at least four dif- ASW. 12 days after the initial inoculations 1 ml of each ferences within the most variable positions of their ITS1 strain's culture was inoculated as follows: 100% ASW into sequences. A degree of subjectivity regarding these deci- 50% ASW and vice versa, to test whether viable forms of sions was allowed, depending on the degree of variability each strain persisted in 100% ASW even when they could of the exact region of the alignment in which they not be seen, and in the second case, to test whether strains occurred. However, these decisions always erred on the were growing (or at least moving around) in 50% ASW side of stringency and it is likely that we have underesti- had acclimatized to the degree to which they could persist mated the total number of ITS-types detected in this study. in 100% ASW. Observations reported in Table 2 were If so, this would mean that on average, we overestimate the made from these experimental cultures and others of the degree of cosmopolitanism of each ITS-type. Single nucle- same strains examined over the previous 18 months. Cul- otide differences in conserved regions of the alignment tures used for these experiments are marked with open cir- were discounted as likely sequencing errors and/or poly- cles in Fig. 1. No cultures from clade β were used as they morphic sites. One example of each unique ITS-type pro- are all dead. duced by this study has been deposited in GenBank (Accession numbers EF174336-EF174385). Authors' contributions DB coordinated the study, generated the dataset for group PCR recombination (chimaera formation) α, supervised the generation of datasets for β and γ, and Chimaeric sequences are caused by DNA template switch- was the main drafting author. DB, TAR, and TCS collected ing between PCR cycles. We took the following precau- samples. LM and VM generated the datasets and executed tions to eliminate chimaeras from our dataset. 1) Our preliminary analyses for groups β and γ, respectively. TAR amplicons were as short as possible, and PCR extension contributed scientific discussion, helped analyze the raw times were set for at least three times as long as necessary datasets, generated the phylogenies, created the figures, (this was determined empirically). Chimaera formation and assisted with the MS. TC-S gave general advice and frequency is relatively low in each experiment [10] and is inspiration and helped with scientific evaluation and writ- minimal where amplicon length is small and extension ing. time is long. 2) The short ITS1 region was scanned in order to check that the sequence variability and signature Acknowledgements sequences were consistent along its length. If a PCR This work was supported by a NERC research grant to TCS, which sup- recombination event had occurred it would be easily ported DB; TAR and VM are supported by the BBSRC. TAR and DB thank detectable. 3) Identical sequences recovered from more the Amazon Conservation Association for financial support for collection than one library were deemed non-chimaeric. The likeli- of the Peruvian samples and the Smithsonian Tropical Research Institute, Panama for assistance while collecting and processing the Panamanian sam- hood of the exact same chimaera being produced in inde- ples and cultures. TCS thanks NERC and the Canadian Institute for

Page 12 of 13 (page number not for citation purposes) BMC Evolutionary Biology 2007, 7:162 http://www.biomedcentral.com/1471-2148/7/162

Advanced Research Evolutionary Biology Programme for Fellowship sup- microbacteria affiliated with the Polynucleobacter cluster. port. Appl Environ Microbiol 2004, 70(10):5787-5793. 22. Mitchell EAD, Meisterfeld C: Taxonomic confusion blurs the debate on cosmopolitanism versus local endemism of free- References living protists. Protist 2005, 156:263-267. 1. Baas-Becking LGM: Geobiologie of Inleiding tot de Milieuku- 23. Chao A, Li PC, Agatha S, Foissner W: A statistical approach to nde. In The Hague Edited by: van Stockum WP, Zoon NV. The Neth- estimate soil ciliate diversity and distribution based on data erlands ; 1934. from five continents. Oikos 2006, 114:479-493. 2. Hughes Martiny JB, Bohannan BJM, Brown JH, Colwell RK, Fuhrmann 24. Foissner W: Protist diversity: estimates of the near-imponder- JA, Green JL, Horner-Devine MC, Kane M, Krumins JA, Kuske CR, able. Protist 1999, 150:363-368. Morin PJ, Naeem S, Øvreås L, Reysenback AL, Smith VH, Staley JT: 25. Foissner W: Biogeography and dispersal of micro-organisms: Microbial biogeography: putting microorganisms on the a review emphasizing protists. Acta Protozool 2006, 45:111-136. map. Nat Rev Microbiol 2006, 4:102-112. 26. Sandon H: The composition and distribution of the protozoan 3. Finlay BJ, Fenchel T: Divergent perspectives on protist species fauna of the soil. Edinburgh , Oliver & Boyd; 1927. richness. Protist 1999, 150(3):229-233. 27. Foissner W: Diversity and ecology of soil flagellates. In The Biol- 4. Richards TA, Bass D: Molecular screening of free-living micro- ogy of Free-living Heterotrophic Flagellates Volume 45. Edited by: Patter- bial eukaryotes: diversity and distribution using a meta-anal- son JP, Larsen J. Oxford , The Systematics Association; 1991:93-112. ysis. Curr Opin Microbiol 2005, 8(3):240-252. 28. Ekelund F, Patterson DJ: Some heterotrophic flagellates from 5. Finlay BJ: Global dispersal of free-living microbial eukaryote cultivated soil in Australia. Arch Protistenkd 1997, 148:461-478. species. Science 2002, 296(5570):1061-1063. 29. Cracraft J: Patterns of diversification within continental 6. Foissner W, Strüder-Kypke M, van der Staay GWM, Moon-van der biotas: hierarchical congruence among areas of endemism of Staay SY, Hackstein JHP: Endemic ciliates (Protozoa, Cilio- Australian vertebrates. Aust Syst Bot 1991, 4:211-227. phora) from tank bromeliads (Bromeliaceae): a combined 30. Crisp MD, Laffan S, Linder HP, Monro A: Endemism in the Aus- morphological, molecular, and ecological study. Europ J Protis- tralian flora. J Biogeogr 2001, 28:183-198. tol 2003, 39:365-372. 31. Ekelund F, Daugbjerg N, Fredslund L: Phylogeny of Heteromita, 7. Darling KF, Kucera M, Pudsey CJ, Wade CM: Molecular evidence Cercomonas and Thaumatomonas based on SSU rDNA links cryptic diversification in polar planktonic protists to sequences, including the description of Neocercomonas jut- Quaternary climate dynamics. Proc Natl Acad Sci USA 2004, landica sp. nov., gen. nov. Europ J Protistol 2004, 40:119-135. 101(20):7657-7662. 32. Coleman AW: The significance of a coincidence between evo- 8. Telford RJ, Vandvik V, Birks HJ: Dispersal limitations matter for lutionary landmarks found in mating affinity and a DNA microbial morphospecies. Science 2006, 312(5776):1015. sequence. Protist 2000, 151(1):1-9. 9. Finlay BJ, Fenchel T: Cosmopolitan metapopulations of free-liv- 33. Coleman AW: Paramecium aurelia revisited. J Eukaryot Microbiol ing microbial eukayotes. Protist 2004, 155:237-244. 2005, 52(1):68-77. 10. Bass D, Cavalier-Smith T: Phylum-specific environmental DNA 34. Amato A, Kooistra WH, Ghiron JH, Mann DG, Proschold T, Montre- analysis reveals remarkably high global biodiversity of Cer- sor M: Reproductive isolation among sympatric cryptic spe- cozoa (Protozoa). Int J Syst Evol Microbiol 2004, 54(Pt cies in marine diatoms. Protist 2007, 158(2):193-207. 6):2393-2404. 35. Fokin SI, Przybos E, Chivilev SM, Beier CL, Horn M, Skotarczak B, 11. von der Heyden S, Chao EE, Cavalier-Smith T: Genetic diversity of Wodecka B, Fujishima M: Morphological and molecular investi- goniomonads: an ancient divergence between marine and gations of Paramecium schewiakoffi sp. nov. (Ciliophora, Oli- freshwater species. Eur J Phycol 2004, 39:343-350. gohymenophorea) and current status of distribution and 12. von der Heyden S, Cavalier-Smith T: Culturing and environmen- taxonomy of Paramecium spp. Europ J Protist 2004, 40:225-243. tal DNA sequencing uncover hidden kinetoplastid biodiver- 36. Przybos E: Recent data on the occurrence of species of the sity and a major marine clade within ancestrally freshwater Paramecium aurelia complex in Europe. Folia Biol (Krakow) 2005, Neobodo designis. Int J Syst Evol Microbiol 2005, 55(Pt 53:61-63. 6):2605-2621. 37. Przybos E, Fokin SI: Data on the occurrence of species of the 13. Slapeta J, Lopez-Garcia P, Moreira D: Global dispersal and ancient Paramecium aurelia complex world-wide. Protistology 2000, cryptic species in the smallest marine eukaryotes. Mol Biol 1:179-184. Evol 2006, 23(1):23-29. 38. Rambaut A: Se-Al: Sequence Alignment Editor. 1996 [http:// 14. Karpov SA, Bass D, Mylnikov AP, Cavalier-Smith T: Molecular phy- tree.bio.ed.ac.uk/software/seal/]. logeny of Cercomonadidae and kinetid patterns of Cer- 39. Swofford DL: PAUP*. Phylogenetic Analysis Using Parsimony comonas and Eocercomonas gen. nov. (Cercomonadida, (*and Other Methods), Version 4. Sunderland, Massachusetts , Cercozoa). Protist 2006, 157(2):125-158. Sinauer Associates; 2002. 15. Koch TA, Ekelund F: Strains of the heterotrophic flagellate 40. Yang Z: Among-site rate variation and its impact on phyloge- Bodo designis from different environments vary considerably netic analyses. Trends Ecol Evol 1996, 9:367-372. with respect to salinity preference and SSU rRNA gene com- 41. Wuyts J, De Rijk P, Van de Peer Y, Pison G, Rousseeuw P, De position. Protist 2005, 156(1):97-112. Wachter R: Comparative analysis of more than 3000 16. Moreira D, Lopez-Garcia P: The molecular ecology of microbial sequences reveals the existence of two pseudoknots in area eukaryotes unveils a hidden world. Trends Microbiol 2002, V4 of eukaryotic small subunit ribosomal RNA. Nucleic Acids 10(1):31-38. Res 2000, 28(23):4698-4708. 17. Finlay BJ: Protist taxonomy: an ecological perspective. Phil Trans Roy Soc Lond B 2004, 359:599-610. 18. Boenigk J, Pfandl K, Stadler P, Chatzinotas A: High diversity of the ‘Spumella-like’ flagellates: an investigation based on the SSU rRNA gene sequences of isolates from habitats located in six different geographic regions. Env Microbiol 2005, 7:685-967. 19. Leclerc MC, Barriel V, Lecointre G, de Reviers B: Low divergence in rDNA ITS sequences among five species of Fucus (Phaeo- phyceae) suggests a very recent radiation. J Mol Evol 1998, 46(1):115-120. 20. Logares R, Rengefors K, Kremp A, Shalchian-Tabrizi K, Boltovskoy A, Tengs T, Shurtleff A, Klaveness D: Phenotypically different microalgal morphospecies with identical ribosomal DNA: a case of rapid adaptive evolution? Microb Ecol 2007, 53(4):549-561. 21. Boenigk J, Stadler P, Wiedlroither A, Hahn MW: Strain-specific dif- ferences in the grazing sensitivities of closely related ultra-

Page 13 of 13 (page number not for citation purposes)